1. Field of Invention
The present invention relates generally to guaranteeing appropriate quality of service to virtual processes executing on multitasking operating systems, and specifically in one embodiment to guaranteeing quality of service to virtual hosts servicing client requests to multiple network addresses on a single physical host computer.
2. Background of Invention
With the popularity and success of the Internet, server technologies are of great commercial importance today. Typically, a server program executing on a single physical host computer services client requests made to a single network address allocated to the host. However, using Transmission Control Protocol (TCP) and other transport protocols, a server application executing on a single physical host can be programmed to process requests made to multiple network addresses. Such functionality is known as virtual hosting.
To enable virtual hosting, more than one network address is assigned to a single physical host computer. A server program executing on the host opens a communication transport channel (socket) and allows receipt of incoming communications targeted for any of the multiple network addresses assigned to the host. Accepting a communication request by a server executing TCP is a three step process that comprises waiting for the communication request from a client, sending an acknowledgment signal to the client, and receiving a return acknowledgment signal from the client. This three step process is called “three way handshaking” and is a feature of TCP communication. While accepting a communication request, a server is unable to execute other tasks. Thus, a server cannot wait for one incoming communication request and simultaneously service a separate communication request. However, multiple communication requests made to a plurality of network addresses associated with a single physical host require simultaneous service. For this reason, a virtual host server typically accepts the communications requests itself and creates child processes to service the requests.
Two known methods exist for utilizing child processes to service communication requests. The most common involves accepting the communication request by the server, and then generating (forking) a child process to service it. While the child process services the request made by the client and transmits it to the client, the server is free to accept the next communication request, perhaps from another client. As soon as the server accepts a request, the server generates a new child process, which services the request and then terminates. This method of virtual hosting, know as “fork after accept,” is widely used today.
Another method of virtual hosting comprises generating, by the server, a plurality of child processes. The number of child processes created reflects a desired maximum number of communication requests to service simultaneously. Each child process proceeds to accept a communication request. Thus, each child waits for an incoming communication request, services it, and proceeds to wait for a next request. While a child process is servicing a request it cannot be waiting for another request, but this is allowable because other child processes are waiting for incoming requests. Where all the child processes are busy servicing requests, the maximum desired number of requests is being serviced, and no more can be accepted until one of the child processes finishes servicing a request and begins waiting for a new one. This method, called “fork before accept,” is known and used today, although less commonly than “fork after accept.”
Both virtual host systems that utilize the fork before accept method and those that utilize the fork after accept method can be name-based as well as address-based. As described above, in an address-based virtual host system, each virtual host is identified by a separate network address assigned to a single, physical host. In contrast, in a name-based virtual host system, each virtual host is identified not by a network address, but instead by a domain name. Multiple domain names can be assigned to a single address. Thus, multiple virtual hosts, each identified by a unique domain name, can all be assigned to a single network address. The single network address to which the multiple domain names are assigned is assigned to the single, physical host. Communication requests to a name-based virtual host are made to a domain name, not to a network address. As with address-based virtual host systems, all virtual hosts map to a single physical host.
A virtual host system simulates multiple hosts by servicing client requests made to any of the multiple network addresses or domain names. This is desirable, because providing a unique physical host for each network address or domain name is expensive and inefficient. Hosting services are often provided commercially by an Internet Service Provider (ISP). Without virtual hosting, an ISP would have to provide a separate physical host computer with a unique network address for every customer that purchases host services. Often, a customer purchasing host services from an ISP will neither require nor be amenable to paying for use of an entire host computer. Generally, only a fraction of the processing power, storage, and other resources of a host computer will be required to meet the needs of an individual customer.
Virtual hosting allows an ISP to utilize one physical host computer to provide commercial host services to multiple customers, thereby creating a virtual host server. Each customer is assigned a network address (or domain name), and is provided with resources on the single, physical host computer, effectively sharing the host with other customers. A client computer requests data from a specific customer's host by targeting communication requests to the appropriate network address (or domain name). By utilizing the fork before accept method or the fork after accept method, the virtual host server can service requests to multiple network addresses or domain names. Thus, the functionality of numerous hosts is provided by a single physical host computer, servicing requests made to a plurality of network addresses and domain names by multiple customers.
One shortcoming with virtual hosting as it exists today is the inability to allocate appropriate amounts of computer resources of the physical host to servicing client requests made to specific virtual hosts, and hence to specific customers. Where an ISP provides host services to multiple customers on a single physical computer, it is desirable to allot to each virtual host a specific amount of computer resources appropriate to the needs of the customer, and preferably based upon the amount paid for the services. For example, suppose two customers purchase host services from an ISP. The first customer is a large corporation providing financial services to thousands of clients internationally. The financial services host requires a great deal of storage space, as well as prompt response time to all client requests. Of course, the first customer is willing to compensate the ISP appropriately for providing such a level of host services. The second customer is a sole proprietorship that sells floral arrangements locally. The second customer has a very limited budget, but only requires minimal computer resources. Clearly, it is desirable for the ISP to allocate different percentages of the system resources to the two separate virtual hosts provided by the ISP for the two separate customers. However, this is not possible with conventional virtual hosting techniques.
Multitasking operating systems such as UNIX facilitate specification of resource allocation to multiple concurrent processes. The operating system can be instructed as to how to allocate resources to different processes. System resources can be allocated to processes as a percentage of resources available (for example, the operating system may be instructed to allocate twenty percent of the central processing unit cycles to process A and two percent to process B), or as specific numbers of units (for example, the operating system may be instructed to allocate X cycles per second to process A and Y cycles per second to process B). Such specification of resource allocation is called a guarantee of quality of service.
A server, which is a process, executing on a dedicated physical host services client requests for a single network address (physical host) only. Thus, quality of service can simply be set for the server to the quality of service appropriate for the host. A virtual host server services numerous client requests for multiple virtual hosts. A single virtual host server provides host services for a plurality of customers all of whom may require different quality of service. Although it would be possible to set a single quality of service for the virtual host server, no single quality of service is appropriate for all of the virtual hosts. Because the appropriate quality of service for different virtual hosts is different, providing a single quality of service for all virtual hosts is undesirable. As detailed above, ISP's utilize single server application programs to provide virtual hosting services to multiple customers with varying business needs and budgets. It would be desirable for an ISP to be able to make appropriate quality of service guarantees to different customers purchasing virtual host services. However, conventional virtual host systems can provide only the same quality of service for all virtual hosts supported by the server. With existing virtual host systems, an ISP cannot provide one quality of service to the international financial services corporation of the example given above, and another to the local florist. The ISP either must provide each customer with the same quality of service, which is undesirable, or provide each customer with a separate physical host, which is inefficient and expensive for the ISP, as many customers do not require the resources of a dedicated host computer. What is needed is a virtual host system that provides appropriate quality of service guarantees for each virtual host.
Virtual host systems are being utilized today by ISP's and other providers of host services. As explained above, it would be desirable for existing ISP's providing virtual host systems to be able to provide appropriate quality of service guarantees to multiple virtual hosts. However, many such ISP's would not want to replace their existing virtual host system with one that could provide such guarantees, even if such a system were available. Upgrading a virtual host system is a time consuming and complicated process, often involving costly down time and high labor expenses. It would be desirable to have a system to allow existing users to provide appropriate quality of service guarantees to multiple virtual hosts, without having to upgrade or replace their existing virtual host systems.